This type of ethics aims to protect AI systems from external attacks and improves security for end-users.